As Global Powers Explore Humanoid Robots, Cyber-Risk Looms

In contrast to abstract large language models (LLMs) and artificial intelligence (AI) agents, global superpowers are making significant strides towards a future predicated on embodied AI, utilizing cyberattacks as a strategy to gain an advantage.

Many experts assert that we are on the brink of a new industrial revolution, akin to historical shifts driven by manufacturing, steam power, railroads, electricity, and more recently, the Internet and telecommunications. This revolution is anticipated to be spearheaded by embodied AI systems—robots that can move and operate similarly to humans or animals.

As both corporations and nation states vie for supremacy in the field of intelligent robotics, novel cyber battlegrounds and risks are emerging. At the upcoming Infosecurity Europe event, Joseph Rooke from Recorded Future will present a cybersecurity-focused adaptation of his well-received discourse on the power dynamics, supply chain vulnerabilities, and potential cyberattack scenarios associated with embodied AI systems, applicable both now and in the future.

"The race is on," he notes, "and the current security posture of many of these systems is highly concerning."

Cyber-Risks in Embodied AI

With advancements in kinematics keeping pace with other AI technologies, experts forecast a surge in the deployment of embodied systems. A report from Morgan Stanley last year projected that by 2050, China could have approximately 300 million humanoid robots actively functioning in industrial settings, military units, and other beneficial applications.

Investing in humanoid robots may be a proactive response to demographic declines, or alternatively, a strategy to substitute human workers with machines that operate without wages. Regardless, progress in this domain is well underway. Humanoid robots have already played a significant role in initiatives led by the Chinese Communist Party (CCP), drones have demonstrated critical utility in the Russia-Ukraine conflict, and consumer-grade robot dogs are now available online for a few thousand dollars.

The scope of potential risks associated with embodied AI systems is expansive. Robots capable of physical movement pose evident safety risks to their surrounding environments. Furthermore, robots integrated into homes and industrial spaces can compromise data security, not only via internet connections but also through their sensory capabilities. These systems are also vulnerable to hacking through the same avenues.

The existing problems with currently available robots only amplify concerns. Researchers focusing on this area have extensively examined Unitree bots, which are the preeminent humanoids in the market, revealing that these machines transmit private user data to China without obtaining consent. They have identified multiple backdoors that allow for complete, unauthorized control and demonstrated the potential for rapid exploitation or wireless propagation from one bot to another. "That was genuinely alarming—more like iRobot scenarios. A fleet could be compromised under these circumstances," Rooke recalls.

As with LLMs and agentic AI, companies are hastily launching embodied AI solutions without thoroughly addressing the associated risks, Rooke observes. "It is indeed a frenetic race, and I have concerns about potential oversights. My worry now pertains to the US robotics sector: Are they overlooking critical issues?"

Chinese Spying in the Mining Sector

The cybersecurity concerns surrounding intelligent robots extend well beyond the robots themselves. The supply chains critical for constructing embodied AI systems are evolving and are increasingly vulnerable to commercial and geopolitical cyberespionage.

Humanoid robots face a multitude of cyber threats targeting the AI models they utilize, the data centers they depend on, the semiconductors that power them, and the energy sources implicated throughout their production. Rooke specifically emphasizes a disturbing rise in cyberattacks directed at the global mining sector.

Why the focus on mining? As the demand for embodied AI and its components escalates, the need for rare earth elements and other essential minerals will likewise surge. Control over these natural resources will confer tremendous power within the framework of the fourth industrial revolution, particularly as many rich deposits are found in regions with ambiguous legal status, such as the Arctic and outer space.

China's advanced persistent threat (APT) groups have been notably active in this arena. They've engaged in cyberespionage to gain intelligence on foreign strategic initiatives, undermine rival plans, or appropriate their mining technologies. Recorded Future has traced several mining-specific Chinese cyberespionage campaigns over the past five years along with numerous others that are tangentially related to mining efforts.

• In 2021, APT15 (also known as Nickel or Nylon Typhoon) targeted a Canadian mining entity.
• In 2025, coinciding with China's forays into seabed exploration and mining partnerships with several smaller nations, multiple Chinese APTs were reported spying on an organization focused on regulating seabed mining.
• Between 2021 and 2026, Chinese cyber actors targeted both private and public sector organizations in Indonesia, where China maintains numerous contracts and interests, particularly relating to nickel.
• In 2025, YoroTrooper (also known as Silent Lynx), believed to be based in Kazakhstan, targeted Russia’s energy production, manufacturing, and mining industries.

When a mining contract becomes available, "they may infiltrate a ministerial network or confirm reconnaissance phases: How might we undermine this bid? Their strategies revolve around acquiring an advantage," Rooke explains.

Fortunately, he adds, "Currently, I wouldn’t categorize this as destructive; it primarily entails espionage."